An educational crypto, finance, and AI vocabulary puzzle about market risk, proof-of-stake, model controls, and data quality.
Risk Metrics
These terms track dispersion, co-movement, downside moves, and comparison against a reference.
VolatilityCorrelationDrawdownBenchmark
- Volatility: Volatility measures how much an asset's price moves over time.
- Correlation: Correlation measures how closely two assets or variables move together.
- Drawdown: Drawdown measures the decline from a recent peak to a later low.
- Benchmark: A benchmark is a standard used to compare performance against a market or portfolio.
Consensus Roles
These terms describe who proposes blocks, who validates them, and how time is divided on chain.
ValidatorProposerSlotEpoch
- Validator: A validator checks and helps secure blocks in a proof-of-stake network.
- Proposer: A proposer is the validator selected to suggest a new block.
- Slot: A slot is a short time interval used to schedule block production.
- Epoch: An epoch is a larger bundle of slots used for coordination and accounting.
Inference Controls
These terms cover limits on output length, sampling style, and when generation stops.
Token BudgetTop-PTop-KStop Sequence
- Token Budget: A token budget caps how much text or code a model can generate or process.
- Top-P: Top-p sampling keeps the smallest set of tokens whose cumulative probability reaches a chosen threshold.
- Top-K: Top-k sampling limits selection to the k most likely next tokens.
- Stop Sequence: A stop sequence tells a model where to halt generation.
Data Hygiene
These terms describe checks, structure, labels, and saved versions that make data easier to trust.
ChecksumTimestampSchemaManifest
- Checksum: A checksum is a compact value used to detect accidental changes in data.
- Timestamp: A timestamp marks when an event or record was created or updated.
- Schema: A schema defines the shape and fields of structured data.
- Manifest: A manifest is a file that lists contents, versions, or metadata for a package or dataset.